Why most buyer research changes nothing
Teams interview customers, hear encouraging things, produce a summary document, and change nothing. This is so common it is almost the default outcome.
The cause is nearly always question design. Ask people what they want and they will tell you, fluently and inaccurately. Ask what features would make them buy and you will get a wish list that predicts nothing. Ask how they would feel about a new positioning and you will get politeness.
Research changes strategy when it surfaces what people actually did: how they discovered the problem, who they consulted, what they searched, what nearly stopped them, and why they chose what they chose. Behavior is evidence. Stated preference is conversation.
Recruiting the right people
Three groups tell you different things.
Current customers explain why you won and what nearly went wrong. They are easiest to recruit and the most generous, which also makes them the least reliable on your weaknesses.
Lost prospects are the most valuable and the hardest to book. They know exactly why they chose someone else, and they have no reason to soften it.
Churned accounts explain where the value proposition broke down after the sale, which is often a different story from the one sales tells.
For recruitment, keep the ask small and specific. Twenty minutes, not an hour. Make clear nobody is selling. A modest incentive helps with lost prospects and is unnecessary for engaged customers. Expect roughly one in four to agree among customers and one in ten among lost deals.
Question design that surfaces truth
Four principles do most of the work.
Past behavior over hypothetical preference. Replace what would make you switch with walk me through the last time you evaluated an alternative.
Specific over general. Replace how do you research vendors with what did you search, and what did you open first.
Silence over prompting. After an answer, wait. The second thing people say is usually more honest than the first.
Never defend. When someone criticizes your product, the correct response is to ask them to say more. Any defense ends the useful part of the conversation.
Running the conversation
Open by setting expectations: how long, what for, that there are no wrong answers, and that you would like to record for accuracy. Get explicit consent.
Spend the first few minutes on context, then move to the specific purchase or evaluation. Follow tangents when they seem to matter, because the unplanned detour is frequently where the useful material sits.
Close by asking what you should have asked and did not. This single question regularly produces the most valuable answer in the session.
From transcripts to decisions
Read everything before coding anything. Then tag by theme, noting frequency and intensity separately, because something mentioned by three people with real feeling often matters more than something mentioned in passing by eight.
The threshold for action is not statistical. It is whether the pattern explains something you previously could not explain. When a finding accounts for a conversion gap you had been guessing about, that is enough to act on.
Route findings deliberately: language into messaging, objections into sales enablement, discovery behavior into channel strategy, and evaluation criteria into competitive positioning. Research that stops at a summary deck is where most of this work dies.
The interview guide
The working guide we use is included below, structured by section with timing. Adapt the specifics, keep the sequence.
- Context, 3 minutes: What is your role? What changed before you began looking?
- Trigger, 4 minutes: Walk me through the moment you decided to act. What happened next?
- Evaluation, 6 minutes: Which alternatives did you consider? Who was involved? What evidence mattered? What nearly stopped the decision?
- Choice and experience, 5 minutes: Why did you choose this option? What has worked or disappointed you since?
- Close, 2 minutes: What should I have asked? May we follow up to clarify your answers?
Record the buyer’s words, the evidence behind each finding and the decision it should change. Ask permission before recording.
Common questions
How many interviews before patterns emerge?
Usually eight to twelve within one segment. If you are hearing nothing new by twelve, you have saturation.
Should we interview our own customers ourselves?
You will get useful material, but people are measurably more candid with a third party, particularly about price and about your weaknesses.
Will lost prospects really talk to us?
More often than teams expect, especially if you make clear you are not selling and you keep it to twenty minutes.
Do surveys work instead?
Surveys confirm hypotheses at scale. They do not generate them. Interview first, then survey to size what you found.
